Artificial Intelligence-Guided Optimization of Hyaluronic Acid-Coated Liposomal Linagliptin for Targeted Management of Polycystic Ovary Syndrome

Abstract

Linagliptin, a DPP-4 inhibitor commonly used in the management of diabetes mellitus, has shown potential activity in polycystic ovary syndrome (PCOS). Linagliptin's therapeutic effectiveness is limited by its poor membrane permeability and low oral bioavailability. This study aimed to formulate hyaluronic acid-coated liposomal linagliptin optimized through I-optimal design and AI-based entrapment efficiency (EE%) prediction. The effects of hyaluronic acid and drug concentrations on particle size (PS), polydispersity index (PDI), zeta potential (ZP), and EE% were systematically evaluated to develop an optimized delivery system for PCOS management. The optimized formulation (O₁) demonstrated a PS of 152.5 nm, PDI of 0.373, ZP of -19.92 mV, and an EE% of 89.43%. The integrated AI-based predictive model achieved 89.8% accuracy, confirming its reliability for rational formulation design. ​In-vitro dissolution studies revealed a sustained drug release over 72 h from O₁, in contrast to complete release within 3 h from unformulated linagliptin. In the PCOS-induced rat model, treatment with both unformulated linagliptin and O₁ significantly improved insulin sensitivity and normalized lipid profiles. Notably, O₁ markedly restored ovarian redox balance through modulation of the Keap1/Nrf2 pathway, indicating a mechanistic basis for the amelioration of PCOS-associated oxidative stress and metabolic dysfunction. Overall, the optimized HA-coated liposomal formulation demonstrated superior therapeutic efficacy and bioavailability compared to unformulated linagliptin, supporting its potential as a targeted repurposed nanocarrier-based therapy for PCOS management, where AI and response surface design are efficient tools for accelerating pharmaceutical formulation development and predicting formulation performance.